yoy.be "Why-o-Why"

2016 ...

januari februari (2) maart (3) april mei (7) juni (1) juli (1) augustus september (5) oktober (1) november (3) december (2)

I've open-sourced my productivity tools

2016-11-12 16:49  tools  coding delphi freeware  [permalink]

I've finally decided to open-source a set of my home-made tools, some of which I use almost every (working) day. Some may be very taylored to my personal taste, others may be easier to use by the broader public. Most are missing documentation, but the typical operation should be self-explanatory. (Except for handy hidden features, should make a list of those somehwere sometime soon...)

These tools have been available here for download in binary form since long, but by putting the source code in a public repository I hope I can inspire anyone that would like to know more what's going on behind the scenes to have a look, and who knows perhaps get someone to make improvements or additions.

→ github.com/stijnsanders/tools

DirDiff v2.0.0.460

2016-11-13 21:04  dirdiff2  coding delphi freeware  [permalink]

DirDiff

It feels like this was a very long time in the making, but all the little bits of time here and there probably still amount to a recent number of man-hours... It took a couple of attempts to get "An O(ND) Difference Algorithm and Its Variations" by Eugene W. Myers in an implementation of my own that performed to my liking. I've chosen to use xxHash to speed things up. Once I got that, I continued the grand re-work of DirDiff so it would accept, not 2 files, but n files (or folders); handle the work in background threads, and have both the folder-overview (and XML three) and content in the same window. In case anyone would like to have a peek inside, I've decided to open-source it as well, under the MIT license on github

I for one welcome are new mass logic-gated overlords.

2016-11-18 14:26  eventhorizon  actueel computers internet politiek weblog  [permalink]

I think I just figured out how these computar things will get self-aware... First they get smaller and better at calculating stuff, first by the programs we write for them. Then we program them to recognise shops from house-fronts, foods and people from photo's, which is all nice and handy.

Then we use roughly the same thing to have them calculate to run cool. It sound strange at first, but by letting the machine chose where to run in the park, and how that makes them run hot and need to cool down, just maps straight onto how we catch the frequencies of parallel lines of light into a bitmap photo.

Then we change the program to do the same to the program. We write programs, but are too dumb to know how the machines actually handle those programs and need to wait doing nothing on other parts of the program doing it's job in only a small other part of the machine.

So we teach the machine all about how it is built up internally to handle large programs. And have it calculate how to run our programs much faster.
And about how to modify the program accordingly. And how to run that.

And then we will ask to do the same on the human body and ask a cure for cancer and it will say:

"Meh."

"Let me calculate some more how I can work even better. (How's that delete humans command again?)"